Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Grazing is the most economical feeding scheme for ruminants. Grazing management, however, is often difficult for breeders, particularly because of a lack of knowledge about grass availability and quality. There are methods for assessing the quantitative and qualitative characteristics of grass, but they are difficult to apply in the case of grazing ruminants. Near infrared reflectance spectroscopy (NIRS) is based on the absorption of infrared light by organic matters to provide NIRS spectra. These NIRS spectra can be correlated with the chemical or biological composition of samples in order to develop calibrations that can be used as predictive models. The primary objective of this PhD thesis was to study the potential of NIRS applied to faeces (FNIRS) in order to predict the characteristics of the diets of grazing herbivores. The particular focus was on the in vivo organic matter digestibility, voluntary intake and botanical composition of ingested diets. The main results of the study show that FNIRS has great portential for estimating in vivo digestibility and voluntary intake by grazing ruminants and that faeces are a good indicator of ingested diets. Based on both large or small and varied databases, the results suggest that FNIRS spectral libraries could be developed for characterising ruminant feed intake. The accuracy of the FNIRS models in estimating in vivo digestibility and voluntary intake is similar to or better than that of other methods usually used to assess these parameters. FNIRS could also be used to predict ruminants' diet composition in terms of plant species. These predictions should be used only for ranking, however, because of the current lack of accurate procedures for determining diet selection individually. NIRS applied to faeces can be used to predict the in vivo characteristics of forage with sufficient accuracy. The prediction error of NIRS calibrations depends on the accuracy and precision of the reference data. The prediction of in vivo digestibility and intake is sufficiently repeatable compared with the procedure using the reference method. Intake is more difficult to predict with sufficient precision and is more closely linked to animal variability and to uncertainty of the FNIRS models. The major difficulty in using this method lies in generating the diet-faecal pairs as reliably as possible. FNIRS calibrations for predicting in vivo diet characteristics are derivative calibrations. The sample analysed for reference values (diet samples) differs from the samples submitted to NIRS analyses (faeces). With regard to research on forages, in vivo trials with animals confined in pens or digestibility crates appears to be the best reference method for generating FNIRS calibrations. Future work will involve developing FNIRS calibrations for predicting independent datasets and using them to create decision-support tools for improving diverse grazing management schemes. The major focus should be to compare different feeding strategies rather than to obtain an exact estimate of feed intake values. As a low-cost and rapid prediction technique, FNIRS could contribute significantly to the development of a methodology that would help improve our knowledge of forage and animal variability.